An E-commerce recommendation approach based on collaborative preferences extension clustering

E-commerce recommendation helps consumers to find the products and services they want. Challenging research problems in E-commerce remain. The existing methods tend to use the same theme granularity. However due to the consumer's individual differences and the context of the consumer tasks, different consumers are not possible to understand all the same. Meanwhile, the data sparsity reduces the accuracy of the recommendation system. In this paper, we propose an approach on collaborative preferences extension based E-commerce recommendation that overcomes these drawbacks and try to find the hidden theme preferences, based on the collaborative extension SOM clustering method. We describes our method in three stages: collaborative preferences expansion, preference feature construction, and preferences clustering stage. Experiments show that the proposed approach is effective.